1. Robert Taylor
Mehdi Heris
Austin Troy, PhD
University of Colorado Denver
Presented at the Partners in Community Forestry Conference 2015
How Urban Tree Canopy Regulates Microclimate and Urban Heat
Islands: A Study from Denver and Baltimore

2. Why do we care about urban heat?
http://www.urban-climate-energy.com/urbanHeatIsland.htmEPA

3. How does land cover change impact urban
energy budget?
• Absorptivity, reflectivity, transmissivity, and
emissivity.
• Different surfaces absorb and reflect
different quantities in different wavelengths
of the spectrum. Impervious surface absorbs
lots of radiation in the short-wave portion
• Different surfaces shed heat (emissivity) at
different rates
• Building canyons can impede release of long-
wave energy to atmosphere
• Convective heat transfer
• Anthropogenic heat
• Less evapotranspiration (see next slide)

4. How do trees work with urban heat?
• Evapotranspiration: e.g. 40,000
GPY for large oak. ET cools air by
using heat from air to evaporate
water and can reduce temp.
Effects can be up to 9ºF
• Direct shade (we’ll get to this
later): in summer only 10-30% of
sun’s E reaches area below
• Higher emissivity and less stored
heat

8. impervious surface to canopy ratio

12. North Side
South Side
60 42 8 19 6.7 68 7 76 100
40.72 102 8% 19% 6.7% 68% 7% 0.76 -0.99
Colfax Corridor
Temperature
roof color canopy building area parking area
impervious
surfaces
vacant land
CanImp
index
compactness
index
39.46 107.0 12% 20% 0.8% 51% 1% 0.49 -0.90
40 38 12 20.0 1 51 1 50 90
North Side
South Side
Surface Temperature and Morphology
in Denver

14. Our project: Urban heat at two scales
http://thinkgreendegrees.com/wp-content/uploads/2012/08/urban-heat-island-comparison.gif

15. Research Questions
1. How does the heat island effect differ in a humid temperate city
(Baltimore) versus and semi-arid zone city (Denver)?
2. How does the spatial pattern of tree canopy mediate trees’
influence on the heat island effect and how does this effect vary
between Denver and Baltimore?
3. How can we calculate the amount of tree shade that directly hits
buildings and does the proportion of shaded building area vary
between Denver and Baltimore?

16. Question 1: Heat island in Denver
vs. Baltimore

17. Tree Canopy Comparison
Denver has 31.8 square km of tree canopy coverage Baltimore has 48.4 square km of tree canopy coverage

18. 18
Surface Temperature: Baltimore
From ASTER Satellite

19. Surface Temperature: Denver
Nighttime August 2003 Daytime June 2012

20. Transect analysis
20

21. trend of building area (vertical axis), distance from center (horizontal axis), average
patch area radius of circles), and temperature (color)
21

22. trend of building area (vertical axis), distance from center (horizontal axis), average
patch area radius of circles), and temperature (color)
22

23. Transect analysis
x axis: distance from downtowns
23

24. 24
Denver Baltimore

25. Why?
• In eastern city surrounded by natural forest, outskirts cool down
much quicker at night due to high emissivity relative to city
• In semi-arid location, trees are not endemic to outskirts, rather
urbanization results in MORE trees than would otherwise be there.
• The fact that urbanized areas tend to go along with trees means that
heat trapping effect of impervious area is largely offset by increase in
tree cover relative to surrounding prairie.
• Exceptions: downtown, where tons of building area relative to trees;
airport, where lots of impervious area

26. Question 2: Spatial patterns of
trees and urban heat mitigation

27. Average patch circularity
At 500 m scale:
• Circularity of patch
geometries
cirRatio = (math.pi * 4* area) / math.pow (perimeter,2)
27
Patch circularity(blue is low, red is high)

28. Patch edge/area ratio
At 500 m scale:
• Average edge/area ratio
28
Patch edge area ratio (blue is low, red is high)

29. Patch area/patch envelope ratio
At 500 m scale:
• Average patch area/patch envelope ratio
29
Patch area envelope ratio (blue is low, red is high)

30. Results of Correlations
30
Temperature
5th Jul 2014
Temperature
29th Jun
2012
Temperature
18th Jun 2005
(nighttime)
Temperature
24th Aug
2009
Patch density (number of patches) 0.64 0.57 0.05 0.17
Total patch area -0.39 -0.27 -0.30 0.02
Patch average area -0.28 -0.23 -0.11 -0.03
Patch length 0.39 0.41 -0.12 0.16
Patch circularity 0.38 0.31 0.10 0.12
Patch Length / area ratio 0.44 0.35 0.18 0.11
Patch Envelope Length / width
ratio
0.07 0.08 -0.12 0.08
Patch area to envelope area ratio -0.22 -0.14 -0.24 0.08
Moran's I 0.29 0.34 0.44 0.07
Temperature
5th Jul 2014
Temperature
29th Jun
2012
Temperature
18th Jun
2005
(nighttime)
Temperature
24th Aug
2009
Road area 0.13 0.17 -0.06 0.06
Water area -0.14 -0.25 0.25 -0.01
Grass area -0.04 0.06 -0.36 0.06
Parking area 0.77 0.63 0.31 0.15
Housing unit 2000 0.64 0.56 0.16 0.12
Housing unit 2010 0.64 0.56 0.17 0.12
Building area 0.82 0.71 0.21 0.16
Tree Canopy area -0.47 -0.27 -0.30 0.02
Patch area to
minimum bounding
envelope area ratio
-0.02 0.05 0.01 -0.05
Patch orientation 0.21 0.15 0.07 0.21
Minimum bounding
geometry length to
width ratio
0.02 0.01 0.02 0.02
Gray= Nighttime temperature
Yellow= Relatively high correlation values
Orange= Statistically insignificant correlations

31. Patch correlation trend analysis
Levelling off point around 5000 sq m. 31
we measured correlation of temperature and patch-length/area-ratio
when the patch average area increases

32. Results so far for Baltimore
• “Edginess” has less of an impact on increasing temperatures for small
patches than large patches of forest; influence increases until 5000 square
meter patch size
• Effects of tree canopy pattern when controlling for impervious/ built area
(R-squared ~ .78):
• Patch area decreases temp
• Patch density (means more, small patches) increases temp
• Patch length increases temp
• Patch circularity (associated more with individual crowns) increases temp
• Patch edge to area ratio increases temp
• Circularity and edge-area results may seem at odds, but they proxy
different things
32

33. Question 3: Quantifying tree
shade hitting buildings: Denver
vs. Baltimore

34. Analysis of shade based on LiDAR
6pm Shadow from Buildings & Canopy
34

35. 35

36. Shade Effects – Solar Angle & Azimuth

37. Shade Effects – Integrated over time
9am
6am
12pm
3pm
6pm

38. Shade Effects – Isolated & CombinedTrees Buildings Trees & Buildings

39. 3D Shade Approach: 8am

40. 3D Shade Approach: 9am

41. Methods
1) Data Processing
2) Insolation
Calculations
3) Shade Calculations
4) Insolation-Shade
Integration
Data Processing
Insolation
Calculations
Shade
Calculations
Insolation-
Shade
Calculations
Data Processing
Statistical
Processing
Unclassified
LiDAR Data
Vector Data
DEM Buildings
DEM
Deciduous
DEM
Coniferous
DSM
Sun’s Position
Data
LAI
Solar Potential
Grid Roof
Surface
Shade Effect
Grid Surface
Solar Potential
Post-Shade
Effect
Aggregated
Shade Effect
per Apt. Unit
Resulting
Statistical
Relationships
1
2 3
4
5
41

42. Total Shade Comparison
0
100
200
300
400
500
600
ShadeArea[km^2]
Hours
Denver Shade
Area
Baltimore Shade
Area
Total Tree Shade : 4 hour interval (summed) * 3 days (June 15, July 15, August 15)
0
10
20
30
40
50
60
70
80
90
100
1 2 3 4 5 6 7 8 9 10 11 12
ShadeArea[km^2]
Hours
Denver Shade Area
Baltimore Shade Area

43. Roof-Tree Intersection Shade Comparison
Roof-Tree Intersection Shade : 4 hour interval (summed) * 3 days (June 15, July 15, August 15)
0
2
4
6
8
10
12
1 2 3 4 5 6 7 8 9 10 11 12
ShadeArea[km^2]
Hours
Denver Shade Area
Baltimore Shade Area
0
5
10
15
20
25
30
35
40
45
ShadeArea[km^2]
Hours
Denver Shade
Area
Baltimore Shade
Area

44. Conclusion and next steps
• Heat island effect is very different in eastern and western US
• Western cities surrounded by treeless lands tend to have more trees
relative to natural surroundings than do eastern cities
• Spatial pattern of tree canopy has a big impact on heat mitigation
• Tree shade varies between Denver and Baltimore: more overall tree
shade in Baltimore, but more tree shade hitting buildings in Denver
by far
• Next step: relate this to energy consumption data

45. Thanks to the Baltimore Ecosystem Study and the USDA Forest Service’s Northern Research Station and
Northeastern Area State and Private Forestry Program for their support of this research
Robert Taylor
Mehdi Heris
Austin Troy, PhD
University of Colorado Denver
Austin.troy@ucdenver.edu
Thanks! Questions?